[pca] Patch 142901-05: installation woes

[Glenn Satchell]

On 03/10/10 10:15, Scott A. Severtson wrote:
> Follow-up from the support engineer:
>
>  >> PCA uses patchadd internally to apply patches. Would we be having
>  >> this discussion if we had used PCA to download the patches, then
> applied
>  >> each one manually using patchadd? What if we had written a shell script
>  >> to loop through the patches, and run patchadd?
>  >>
>  >> How would the latter be different than using PCA to apply the patches?
>  >
>  > We also do not provide support assistance for any custom scripts
> which apply
>  > multiple patches; only for the Sun-provided patch cluster
> installation scripts
>  > and even those are provided mostly as a convenience. Any issues with any
>  > such tools must be reproduced using manual installation methods to
> receive
>  > full support.
>
> --- SNIP ---
>
>  > The only methods to *repair* the system and return it to a supportable
>  > configuration are as follows:
>  >
>  > 1) reinstall
>  > 2) upgrade install
>  > 3) revert to a point before PCA was used and return to the current patch
>  > levels without it. If the issue persists we can begin developing a fix as
>  > mentioned above so you can revert once more and then apply the fix.
>  >
>  > In order to assist further, we need you to perform one of the above
> actions.
>  > The issue either will or it will not persist and we can resume the
> investigation
>  > from that point, if necessary, with renewed confidence in the
> integrity of the
>  > rest of the system.
>
> We're going with the "upgrade install" route - don't have the
> time/energy to fight this battle while the server is non-functional.
>
> --Scott
>
I understand your position, and the path of least pain is often the one 
to take.

I just noticed 142901-06 is out. These are the latest fixes, so it 
doesn't look like your problem is included:

Problem Description:

6840801 race between CL_TICK and disp() leaves onproc thread with 
t_waitrq set
6887205 program using POSIX AIO gets unexpected SIGPOLL (SIGIO) signals
6899523 thread stuck in udp_quiesce_conn due to flaw in UDP perimeter 
mechanism
6908371 copyright year needs updating to 2010 for S10U9
6911810 panic: NULL pointer dereference in unix module - page_retire_hunt()
6915007 page_retire_hunt may perform unnecessary memory scans

What about backing out -05 and trying -06, or is the system too messed 
up to be able to do that?

-- 
regards,
-glenn